Explorer — autonomous testing
Explorer turns the AI loose on your application within a strict budget, probing flows nobody scripted. When it finds something — an uncaught exception, a server error, a dead control, a blank page — it writes a ready-to-run test case that reproduces it.
Guardrails
You stay in control of what it may do:
- Forbidden verbs — delete, pay, place order, sign out and similar are vetoed by the engine, not merely discouraged in a prompt.
- Allowed domains — it cannot wander off your application.
- Probe categories — choose which kinds of probing are permitted: navigation, forms, controls, search, pagination, actions.
- Budgets — hard caps on actions, elapsed time and AI engagements.
- Charter — optional free-text guidance steering where it explores. It never overrides the rules above.
Findings arrive with severity, evidence, a screenshot and a drafted test case you can review and add to a suite.